Neural antecedents of self-initiated actions in secondary motor cortex
Murikami, M. Vicente, Gil Costa & Z. Mainen
Nature Neuroscience, vol. 17, No. 11, pp. 1574-82, November 2014
Summary and review of the above paper
INTRODUCTION: In a very guarded fashion, the authors of this paper cast doubt on the Libet-based orthodoxy where pre-conscious activity in the brain precludes the possibility of conscious will. Their research suggests that a simple threshold of neural activity is not the whole story because at any point inputs from the evaluative parts of the brain may alter a subject’s choice.
Decision-making involves the selection of goals or actions with particular timings. Neurons in the frontal and parietal cortex have gradual increases in neural activity in response to visual stimuli. It may be necessary to compromise between the speed and the accuracy of an action, and noisy sensory signals may give rise to variations in response time.
Like waiting for a bus
An important aspect of action and behaviour is the subject deciding when to give up waiting for an event that is expected, but the occurrence and timing of which is uncertain. This decision between either waiting, or giving up, is referred to as inter-temporal. Subjects may have decided to wait for an event, but nevertheless while waiting they may choose something that is more immediately available.
Historically studies have identified activity in the motor, parietal and prefrontal cortices preceding the subjects awareness of the attention to act. In the study described in this paper, a small reward was available to subjects immediately, but a larger reward was available after a longer and randomised delay. Subjects often gave up on waiting for the larger reward, but the timing of this opt out varied greatly, in ways mainly not related to variations in the trials themselves.
With the timescale chosen in this study, about 30% of subjects succeeded in waiting for the larger reward. Those who did not succeed did not usually take the small reward immediately, but instead took it somewhere during the time gap between the small and the large reward. Moreover, the subjects decided to give up on the larger reward without receiving any new external signal, so the decision taking appears to come from within the subject.
The researchers worked to correlate the variability of decisions The variations in timing were found to be correlated with the activity of populations of neurons in the rostral secondary motor cortex. A large section of this brain region is active during this decision taking process, and the build up in neuron activity was completed just before the subject gave up waiting. The activity of motor cortex neurons correlated with the time the subject was found to wait for the larger reward.
The study recorded two set of neurons whose activity correlated with the onset of an action. The first set of neurons showed increasing activity peaking just before movement, with the rate of build of activity inversely proportional to the length of the subject’s waiting time. This form of activity is apparent in the frontal eye field, motor cortices, and lateral intraparietal area.
Value decision by other neurons
A second class of neurons were transiently active throughout the subject’s waiting period and may be located in the posterior parietal, the medial prefrontal and the striatum. These neurons were seen to carry information about the timing of movement initiation, based on perceptual and value decision variables that ‘vote’ for shorter or longer term waiting times.
Pairs of neurons that were correlated with behaviour were also positively correlated with one another, while other neurons were negatively correlated. Some models suggest that the increasing level of neuronal activity observed is due to recurrent excitatory connections. The researchers were not anticipating some of these inter-neuronal correlations and their correlations with behaviour.
It is possible that the integration of decisions involves a network of cortical and subcortical regions connected to the motor cortex. Waiting or responding involved different outcomes and different costs for the subject. Waiting time predictive neurons may be updated in line with experience and come to reflect decision values. The slow dynamics observed in this system is more likely to arise in a larger network. It is thought the system could include the ventrolateral orbitofrontal and the dorsomedial striatum, brain areas involved with evaluation and decision taking.
Investigative starting point
The experiment suggests that more than one input may be involved in the choice of waiting time. A simple threshold of activity cannot fully predict action. At any moment up to activity reaching threshold opposing, inputs may intervene to give a different result. In their closing sentence, the authors remark that their results provide a starting point for investigating concepts such as will and intention to act.